{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:J5AGE4GIQF2GNBIID6IXVE3LII","short_pith_number":"pith:J5AGE4GI","schema_version":"1.0","canonical_sha256":"4f406270c881746685081f917a936b423c573ec24ec739af48937a6a2c20c7d4","source":{"kind":"arxiv","id":"2605.15522","version":1},"attestation_state":"computed","paper":{"title":"Stochastic Non-Smooth Convex Optimization with Unbounded Gradients","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"AdamW with clipped updates outperforms SGD and AdaGrad on convex stochastic problems with unbounded gradients.","cross_cats":["cs.LG"],"primary_cat":"math.OC","authors_text":"Dmitry Kovalev","submitted_at":"2026-05-15T01:43:22Z","abstract_excerpt":"Much of the existing theory on first-order non-smooth optimization is built on a restrictive assumption that the gradients of the objective function are uniformly bounded. We introduce a much more realistic class of generalized Lipschitz functions, where the gradient norms are bounded by an affine function of the optimality gap. We then ask a natural question: what algorithm achieves the best global convergence rates for solving convex stochastic generalized Lipschitz optimization problems? To address this, we develop a new convergence analysis for several existing algorithms and find that Ada"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":true,"formal_links_present":false},"canonical_record":{"source":{"id":"2605.15522","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"math.OC","submitted_at":"2026-05-15T01:43:22Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"36c1edce1ac2cb48ee65483611a8698eddc85430de0bb7b6fffaf5ab624d7fbc","abstract_canon_sha256":"45cce6e35ed59ada4f5a31b882ff6434d169dc2587a8748eb47e178a18505745"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:01:03.044951Z","signature_b64":"ykjhWXeU4rxrr+aC3KKmd9fc7Pp7RQzN1bQ2OZ/yJ4F0tU+b6RyPAFZvF0PfjwFFRmgk5VLsnfBYBtCRNW0OBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4f406270c881746685081f917a936b423c573ec24ec739af48937a6a2c20c7d4","last_reissued_at":"2026-05-20T00:01:03.043859Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:01:03.043859Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Stochastic Non-Smooth Convex Optimization with Unbounded Gradients","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"AdamW with clipped updates outperforms SGD and AdaGrad on convex stochastic problems with unbounded gradients.","cross_cats":["cs.LG"],"primary_cat":"math.OC","authors_text":"Dmitry Kovalev","submitted_at":"2026-05-15T01:43:22Z","abstract_excerpt":"Much of the existing theory on first-order non-smooth optimization is built on a restrictive assumption that the gradients of the objective function are uniformly bounded. We introduce a much more realistic class of generalized Lipschitz functions, where the gradient norms are bounded by an affine function of the optimality gap. We then ask a natural question: what algorithm achieves the best global convergence rates for solving convex stochastic generalized Lipschitz optimization problems? To address this, we develop a new convergence analysis for several existing algorithms and find that Ada"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"AdamW with clipped updates theoretically outperforms other popular stochastic optimization methods, such as SGD and AdaGrad, for convex stochastic generalized Lipschitz optimization problems.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"Objective functions belong to the newly introduced generalized Lipschitz class in which gradient norms are bounded by an affine function of the optimality gap.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Introduces generalized Lipschitz class and shows clipped AdamW outperforms SGD and AdaGrad for stochastic convex optimization under this and related assumptions.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"AdamW with clipped updates outperforms SGD and AdaGrad on convex stochastic problems with unbounded gradients.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"1a11b3dd57969d76bd9f2221a527b49d8480ec234bb8b417cf1cc8a197cc783d"},"source":{"id":"2605.15522","kind":"arxiv","version":1},"verdict":{"id":"1de4c210-3976-49b9-be99-5c1f364ce5b4","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-19T15:11:36.275720Z","strongest_claim":"AdamW with clipped updates theoretically outperforms other popular stochastic optimization methods, such as SGD and AdaGrad, for convex stochastic generalized Lipschitz optimization problems.","one_line_summary":"Introduces generalized Lipschitz class and shows clipped AdamW outperforms SGD and AdaGrad for stochastic convex optimization under this and related assumptions.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"Objective functions belong to the newly introduced generalized Lipschitz class in which gradient norms are bounded by an affine function of the optimality gap.","pith_extraction_headline":"AdamW with clipped updates outperforms SGD and AdaGrad on convex stochastic problems with unbounded gradients."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.15522/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"doi_title_agreement","ran_at":"2026-05-19T15:31:17.690711Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T15:20:34.474700Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"cited_work_retraction","ran_at":"2026-05-19T14:22:03.038115Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"claim_evidence","ran_at":"2026-05-19T14:21:54.045363Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"shingle_duplication","ran_at":"2026-05-19T13:49:41.841841Z","status":"skipped","version":"0.1.0","findings_count":0},{"name":"citation_quote_validity","ran_at":"2026-05-19T13:49:41.379071Z","status":"skipped","version":"0.1.0","findings_count":0},{"name":"ai_meta_artifact","ran_at":"2026-05-19T13:33:22.627385Z","status":"skipped","version":"1.0.0","findings_count":0}],"snapshot_sha256":"a147eb8a3fb6396790e9f6f3f246dacf131b827a878ddc4fdab1fb44c81422f0"},"references":{"count":22,"sample":[{"doi":"","year":null,"title":"Adam: A Method for Stochastic Optimization","work_id":"1910796d-9b52-4683-bf5c-de9632c1028b","ref_index":1,"cited_arxiv_id":"1412.6980","is_internal_anchor":true},{"doi":"","year":null,"title":"Decoupled Weight Decay Regularization","work_id":"07ef7360-d385-4033-83f7-8384a6325204","ref_index":2,"cited_arxiv_id":"1711.05101","is_internal_anchor":true},{"doi":"","year":1905,"title":"Why gradient clipping accelerates training: A theoretical justification for adaptivity","work_id":"fbf2f93c-f73e-413f-b24f-e6153c29bf03","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":1938,"title":"Near-optimal methods for minimizing star-convex functions and beyond","work_id":"c69cc4c3-a26d-40c9-98be-6897c99bae26","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Methods for convex (l_0, l_1)-smooth optimization: Clipping, acceleration, and adaptivity","work_id":"02737a27-7cd5-4b7a-a102-1c2c4d87a916","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":22,"snapshot_sha256":"78d16cf0c3a5c950b97f185aadbb248c40a422e47582e75d3e07f33375deacc9","internal_anchors":5},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2605.15522","created_at":"2026-05-20T00:01:03.044020+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.15522v1","created_at":"2026-05-20T00:01:03.044020+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.15522","created_at":"2026-05-20T00:01:03.044020+00:00"},{"alias_kind":"pith_short_12","alias_value":"J5AGE4GIQF2G","created_at":"2026-05-20T00:01:03.044020+00:00"},{"alias_kind":"pith_short_16","alias_value":"J5AGE4GIQF2GNBII","created_at":"2026-05-20T00:01:03.044020+00:00"},{"alias_kind":"pith_short_8","alias_value":"J5AGE4GI","created_at":"2026-05-20T00:01:03.044020+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/J5AGE4GIQF2GNBIID6IXVE3LII","json":"https://pith.science/pith/J5AGE4GIQF2GNBIID6IXVE3LII.json","graph_json":"https://pith.science/api/pith-number/J5AGE4GIQF2GNBIID6IXVE3LII/graph.json","events_json":"https://pith.science/api/pith-number/J5AGE4GIQF2GNBIID6IXVE3LII/events.json","paper":"https://pith.science/paper/J5AGE4GI"},"agent_actions":{"view_html":"https://pith.science/pith/J5AGE4GIQF2GNBIID6IXVE3LII","download_json":"https://pith.science/pith/J5AGE4GIQF2GNBIID6IXVE3LII.json","view_paper":"https://pith.science/paper/J5AGE4GI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.15522&json=true","fetch_graph":"https://pith.science/api/pith-number/J5AGE4GIQF2GNBIID6IXVE3LII/graph.json","fetch_events":"https://pith.science/api/pith-number/J5AGE4GIQF2GNBIID6IXVE3LII/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/J5AGE4GIQF2GNBIID6IXVE3LII/action/timestamp_anchor","attest_storage":"https://pith.science/pith/J5AGE4GIQF2GNBIID6IXVE3LII/action/storage_attestation","attest_author":"https://pith.science/pith/J5AGE4GIQF2GNBIID6IXVE3LII/action/author_attestation","sign_citation":"https://pith.science/pith/J5AGE4GIQF2GNBIID6IXVE3LII/action/citation_signature","submit_replication":"https://pith.science/pith/J5AGE4GIQF2GNBIID6IXVE3LII/action/replication_record"}},"created_at":"2026-05-20T00:01:03.044020+00:00","updated_at":"2026-05-20T00:01:03.044020+00:00"}